Biology subjects

Knight, J.

Publications and source records attributed to Knight, J..

At least 19 recordsLinked to original sources

bioRxiv2026

ClonoDeconv: reference-based α - β paired recurrent clonotype proportion estimation from bulk-TCR sequencing

T cell receptor (TCR) sequencing is crucial when studying antigen-specific T cell responses and clonal dynamics. A key objective of TCR repertoire analysis is to identify recurring clonotypes across timepoints or different disease states, providing insight into the evolution of immune response. Bulk-TCR sequencing enables high-throughput profiling of TCR and {beta} sequences but lacks pairing information, while single cell TCR sequencing (scTCR-seq) retains pairing information but is limited by sensitivity, throughput, and costs. We present ClonoDeconv, a computational framework to deconvolute paired TCR {beta} ; clonotype frequency from bulk TCR sequencing experiments using a paired scTCR-seq reference. ClonoDeconv robustly estimated shared clonotype abundances from 5' RACE bulk-TCR datasets using a scTCR-seq reference from the same donor. Applied to longitudinal samples, ClonoDeonv successfully captured the expansion and contraction of rare virus-specific clones across timepoints. Taken together, ClonoDeconv provides a cost-effective and scalable approach for studying the longitudinal T-cell clonal dynamics.

bioinformatics↗

bioRxiv2026

Higher-order interdependencies of the epileptogenic network in presurgical resting-state fMRI

Drug-resistant focal epilepsy (DRFE) is understood as a complex brain network disorder. As a result, conventional pair-wise functional connectivity may fail to capture the higher-order interactions between groups of epileptogenic regions. We applied multivariate information theory to stereotactic electroencephalography-defined networks in retrospective resting-state fMRI analysis to determine whether the overall strength and dominant type of interdependencies between three or more regions, represented by the S-information and O-information respectively, could offer clinically relevant insights prior to surgical intervention in DRFE. We found networks involving the epileptogenic zone to be mostly dominated by re-dundant information-sharing, which was predicted by stronger structural and functional connections between regions. Non-temporal lobe epilepsy patients demonstrated greater redundancy and stronger comparative interdependencies than temporal lobe epilepsy patients, in line with a more complex network architecture. Among non-temporal lobe cases with postoperative follow-up, not seizure-free patients displayed more positive O-information deviation scores than seizure-free patients, suggesting elevated redundancy in networks associated with poorer surgical outcomes. Our findings potentiate the S-information and O-information as complementary higher-order descriptors of the epileptogenic network within personalised presurgical assessments.

neuroscience↗

bioRxiv2026

Mode of T cell priming durably shapes the TCR repertoire, effector function and α4β1 integrin expression of human virus-specific CD4+ T cells

The generation of an effective T cell response against an antigen depends on the recognition of the antigen by the T cell receptor (TCR), followed by T cell priming, initiating coordinated biophysical, biochemical and proliferative changes that drive differentiation into effector and memory clones. The immunological environments in which priming occurs, such as natural infection or vaccination, influences the quality and persistence of memory T cells, but the long-term impacts remain incompletely understood. Here, we investigate how the mode of priming shapes durable antigen-specific CD4+ T cell memory, utilising two cohorts 3-4 years after initial antigen encounter: individuals recovered from SARS-CoV-2 infection and infection-naive individuals who received a SARS-CoV-2 vaccination. Using ex vivo single-cell RNA sequencing, paired TCR sequencing and in vitro functional analyses, we characterise the transcriptional, clonal and functional profile of Spike-specific CD4+ T cells. Across both cohorts, CD4+ T cell responses against spike epitopes S166-180, S751-765 and S866-880, were immunodominant, with shared public TCR clonotypes indicating conserved antigen-recognition regardless of mode of priming. Despite this shared specificity, infection-primed individuals exhibited greater TCR repertoire diversity and lower CDR3{beta} sequence convergence. Transcriptionally, infection-primed cells exhibited a more cytotoxic and effector phenotype, while vaccine-primed cells preferentially adopted T follicular helper (Tfh)- and Th1-associated phenotypes. Infection-primed individuals also displayed enrichment of cell adhesion and integrin signalling pathways, with a greater proportion of spike-specific CD4+ T cells expressing 4{beta}1 integrin subunits, consistent with enhanced migratory and effector potential. Collectively, our findings demonstrate that the mode of antigen priming may influence the long-term CD4+ T cell memory states, influencing TCR repertoire diversity, functional differentiation and tissue-homing potential, years after the initial immune response.

immunology↗

bioRxiv2026

Emergent control of ant learning walks using the mismatch between path-integration and visual cues

Despite their small brains, desert ants can safely navigate back to their nest after travelling hundreds of metres to find food. This ability relies on visual memories, gathered on initial Learning Walks (LWs), during which ants slowly explore the surroundings of the nest. LWs follow a clear progression: early walks are short, spiraling, and interspersed with pirouettes: brief stops during which the ant perform a visual scan. Successive LWs become straighter and extend farther from the nest, until they transition into foraging trips. However, when the visual panorama changes, LWs reappear. Existing models explain how stored panoramic views can guide homing, but not when visual memories are collected or how LW dynamics are controlled. Here, we propose that a single error signal - generated from the mismatch between the estimated nest direction provided by path integration and visual predictions - scaffolds visual learning and drives LW dynamics. Using 3D reconstructions of desert ant experiments, we show that this mechanism reproduces the main features of LWs, including the transition to foraging and reoccurrence of LWs after environmental change. Our model is biologically plausible and consistent with known insect navigation circuits. Author summaryA tiny nest entrance in the desert is easy to miss, but doing so may have lethal consequences for a foraging desert ant, returning home after a long scavenging trip. To solve this problem, desert ants spend several days performing Learning walks (LWs) to learn the visual landmarks around their nest before they begin searching for food. LWs are structured active learning procedures and clear patterns can be identified across ant species. LWs take the form of nest-centred spirals, interspersed with frequent visual scans. Over two to four days, LWs become longer, straighter and walking speed increases, before the ant finally transitions to foraging. LWs may reoccur even in experienced foragers, when a new landmark is introduced near the nest. No previous model has provided a mechanism for how LWs are controlled. Here, we propose that the transition from LWs to foraging is driven by visual familiarity. In our model, unfamiliar surroundings promote learning and exploration; as the visual panorama becomes familiar, behaviour shifts to foraging. Using 3D simulations of experimental layouts, we demonstrate that LWs can emerge from simple rules, and that our simulated agents reproduce the main features of LWs observed in desert ants.

animal behavior and cognition↗

bioRxiv2026

LLM-Evolved Regularization Schedules Prevent Posterior Collapse in Latent Factor Analysis via Dynamical Systems

Latent Factor Analysis via Dynamical Systems (LFADS) is a powerful variational autoencoder for inferring neural population dynamics from spike train data. However, LFADS suffers from pos-terior collapse, where the learned posterior collapses to the prior, eliminating meaningful latent representations. Current solutions require computationally expensive Population-Based Training (PBT) to dynamically tune regularization hyperparameters. Here, we demonstrate that Large Lan-guage Model (LLM)-based program evolution can discover regularization schedules that prevent posterior collapse without PBT. Using FunSearch, an evolutionary algorithm that uses LLMs to generate and refine Python functions, we evolved adaptive regularization schedules that respond to training dynamics. Our best evolved schedule prevents posterior collapse across all tested conditions, maintaining KL divergence 6.5x higher than baseline schedules at 50 epochs (n = 10 seeds each, p < 0.001) and stable above 0.09 through 500 epochs across three Neural Latents Benchmark datasets, while preserving reconstruction quality. This work represents the first application of LLM-based program synthesis to variational autoencoder hyperparameter scheduling, offering a computationally efficient alternative to population-based optimization.

neuroscience↗

bioRxiv2025

The effects of morning priming exercise on afternoon physical and cognitive performance in male rugby union players

ObjectiveTo assess the effect of morning resistance exercise on afternoon physical and cognitive performance in male rugby union players. MethodsOn two occasions (randomised, crossover design), 25 male rugby union players completed morning physical (countermovement jump, 40 m linear sprint) and cognitive (rapid visual information processing; spatial working memory; paired associates of learning) assessments. Control (passive rest) or intervention (barbell back squat, 3 x 3 repetitions at 85% of one repetition maximum and barbell squat jump, 5 x 3 repetitions at 40% one repetition maximum) trials were implemented 5.5 h before afternoon assessments. ResultsDifferences between morning and afternoon assessments were found in sprinting performance following intervention for 30 m fly (-0.08 {+/-} 0.13 s, P = 0.001, ES: -0.75) and 40 m (-0.12 {+/-} 0.17 s, P = 0.004, ES: -0.62) relative to control (0.03 {+/-} 0.07 s, 0.01 {+/-} 0.09 s, 30 m fly and 40 m, respectively). Afternoon countermovement jump flight time improved following intervention compared to control (2% vs. 0%, p = 0.012). Delta analysis discovered significant differences from morning to afternoon in intervention (0.385 ms {+/-} 0.053) but not control (0.405 ms {+/-} 0.057). No changes were found for cognitive assessments (P > 0.05). ConclusionMorning lower body resistance exercise, incorporating a combination of high intensity and ballistic exercises, was an effective morning strategy to improve afternoon markers of physical but not cognitive performance in male rugby union players.

physiology↗

bioRxiv2025

Understanding the mechanism of facilitation in hoverfly TSDNs

Many animals use visual motion cues to track and pursue small, fast-moving targets, such as prey or conspecifics. In target-pursuing insects, including dragonflies and hoverflies, Small Target Motion Detector (STMD) neurons are found in the optic lobes and are believed to be presynaptic to Target Selective Descending Neurons (TSDNs) that project to motor command centres. While STMDs respond robustly to target motion - even when displayed against moving backgrounds - TSDN target responses are modulated by background motion. Depending on whether the background motion is syn- or contra-directional to the target motion, the response of the TSDNs is either suppressed or facilitated (amplified). This suggests that TSDNs not only receive input from STMDs but also from neurons sensitive to background motion, but this neural circuit is not clearly understood. To explore the underlying neural mechanisms, we developed three candidate TSDN circuit models - which combine input from bio-plausible STMDs and optic flow-sensitive Lobula Plate Tangential Cells (LPTCs) in different ways - and fitted them to published electrophysiology data from hoverfly TSDNs. We then tested the best-fitting models against new electrophysiological data using different background patterns. We found that the overall best model suggests simple inhibition from LPTCs with the same preferred direction as the STMDs feeding into the TSDN. This parsimonious mechanism can explain the facilitation and suppression of TSDN responses to small targets, may inform similar studies in other animals and open new avenues for bio-inspired robotics. Author summaryMany human sports, including tennis, football, and basketball, rely on the ability to visually detect and respond to the motion of a small, rapidly moving object. Indeed, some sports stars seem to have an uncanny ability to predict the future location of a ball. Similarly, many animals also need to detect the motion of small objects, as these may represent an approaching predator, prey, or a conspecific that needs to be avoided or pursued. Interestingly, while the detection of small visual targets is computationally a hard problem, insects, which have small brains with very modest power requirements, appear to solve this effortlessly. Thus, understanding how target detection is performed in insects can open up avenues for bio-inspired robotics. Male hoverflies naturally engage in complex pursuit behavior, and since they have specialised neurons for target tracking, hoverflies are ideal for understanding the underlying neural circuitry. Here, we combine neural recordings from hoverflies with computational modelling to show how neurons that respond selectively to target motion are combined with neurons that respond to widefield optic-flow to generate behaviourally relevant sensorimotor responses. Importantly, our model does not involve learning and hence suggests that these behaviours can be innate.

neuroscience↗

bioRxiv2024

Lateralised memory networks explain the use of higher-order visual features in navigating insects

Many insects use memories of their visual environment to adaptively drive spatial behaviours. In ants, visual memories are fundamental for navigation, whereby foragers follow long visually guided routes to foraging sites and return to the location of their nest. Whilst we understand the basic visual pathway to the memory centres (Optic Lobes to Mushroom Bodies) involved in the storage of visual information, it is still largely unknown what type of representation of visual scenes underpins view-based navigation in ants. Several experimental studies have shown ants using "higher-order" visual information - that is features extracted across the whole extent of a visual scene - which raises the question as to where these features are computed. One such experimental study showed that ants can use the proportion of a shape experienced left of their visual centre to learn and recapitulate a route, a feature referred to as "fractional position of mass" (FPM). In this work, we use a simple model constrained by the known neuroanatomy and information processing properties of the Mushroom Bodies to explore whether the use of the FPM could be a resulting factor of the bilateral organisation of the insect brain, all the whilst assuming a "retinotopic" view representation. We demonstrate that such bilaterally organised memory models can implicitly encode the FPM learned during training. We find that balancing the "quality" of the memory match across left and right hemispheres allows a trained model to retrieve the FPM defined direction, even when the model is tested with other shapes, as demonstrated by ants. The result is shown to be largely independent of model parameter values, therefore suggesting that some aspects of higher-order processing of a visual scene may be emergent from the structure of the neural circuits, rather than computed in discrete processing modules. Author summaryMany insects are excellent visual navigators, often relying on visual memories to follow long foraging routes and return safely to their nest location. We have a good understanding of the neural substrates supporting the storage of visual memories in ants. However, it is still largely unknown what type of representation of visual scenes underpins the functions of visual navigation. Experimental studies have shown ants using "higher-order" features as part of navigation, that is features that are extracted across the whole extent of a visual scene. Using an anatomically constrained model of the insect memory centers, we address the question of whether the use of higher-order visual features may be emergent from the overall architecture of the vision-to-memory pathways. We find that balancing the quality of left and right visual memory matches provides an explanation for some higher-order visual processing and visual cognition shown in experiments with ants. Overall, this constitutes a contribution to our understanding of visual cognition and the processing of visual scenes used in navigational tasks. We additionally postulate a novel mechanism ants may use to navigate, which is supported by the bilateral structure of the insect brain.

neuroscience↗

bioRxiv2024

The global fish and invertebrate abundance value of mangroves

Mangroves are a critical coastal habitat that provides a suite of ecosystem services and supports livelihoods. We undertake the first global analysis to estimate density and abundance of 37 commercially important fish and invertebrates that are known to extensively use mangroves. Geomorphic mangrove type, sea surface salinity and temperature, and length of mangrove forest edge were important in predicting the density of commercial fish and invertebrates, with deltaic systems supporting the highest densities. The model predicted high densities throughout parts of southeast Asia, the northern coast of South America, the Red Sea, and the Caribbean and Central America. Application of our model onto the global mangrove extent, estimates that mangroves support the annual abundance of nearly 800 billion young-of-year fish and invertebrates contained in our model. Our results confirm the critical role of mangroves globally in supporting fish and fisheries, and further builds the case for their conservation and restoration.

ecology↗

bioRxiv2023

Liquid channels within B. subtilis biofilms allow the escape of trapped clones and population rescue

Phenotypic heterogeneity is one of the hallmarks of the biofilm lifestyle, where even isogenic populations give rise to spatially organized and phenotypically distinct subpopulations. One such pattern is generated by the ability of several biofilm-forming bacteria to switch between a flagellated and a matrix producing state. Here, using Bacillus subtilis as a model system, we investigate the role of this switch during biofilm development on a solid-air interface. By comparing the matrix-flagella spatio-temporal patterns in wild-type biofilms with mixtures of flagella- and matrix-null mutants biofilms, we find that pattern formation does not require a phenotypic switch that enables individual cells to respond to the local environment, but can be explained by a completely stochastic switch coupled to a phenotype-dependent fitness landscape that selects phenotypes at the population level. Integration of experiments and physical models shows that the coexistence between flagellated and matrix-producing cells provides the population with enhanced resilience to environmental changes, by enabling cells to manipulate and harness the local morphological and transport properties within the biofilm. Our results not only reveal a new evolutionary advantage of phenotypic plasticity in biofilms, but also illustrate how the biology and ecology of these populations are intrinsically tied to their physical properties.

evolutionary biology↗

bioRxiv2021

Extinction risk modeling predicts range-wide differences of climate change impact on Karner blue butterfly

Withdrawal StatementThe authors have withdrawn their manuscript owing to the fact that the manuscript has not gone through the US Geological Survey review and approval process required prior to posting and is, therefore, withdrawn. Therefore, the authors do not wish this work to be cited as reference for the project. If you have any questions, please contact the corresponding author.

ecology↗

bioRxiv2021

TREM2-H157Y Increases Soluble TREM2 Production and Reduces Amyloid Pathology

The p.H157Y variant of TREM2 (Triggering Receptor Expressed on Myeloid Cells 2) has been reported to increase Alzheimers disease (AD) risk. This mutation in the extracellular domain of TREM2 localizes at the cleavage site, leading to enhanced shedding. Here, we generated a novel Trem2 H157Y knock-in mouse model to investigate how this H157Y mutation impacts TREM2 proteolytic processing, synaptic function, and AD-related amyloid pathology. Consistent with previous in vitro findings, TREM2-H157Y increases the amount of soluble TREM2 (sTREM2) in the cortex and serum of mutant mice compared to the wild type controls. Interestingly, the Trem2 H157Y variant enhances synaptic plasticity without affecting microglial density and morphology. In the presence of amyloid pathology, TREM2-H157Y surprisingly accelerates A{beta} clearance and reduces amyloid burden and microgliosis. Taken together, our findings support a beneficial effect of the Trem2 H157Y mutation in synaptic function and in mitigating amyloid pathology. Considering the genetic association of TREM2 p.H157Y with AD, we speculate TREM2-H157Y might increase AD risk through an amyloid-independent pathway, as such its effects on tauopathy and neurodegeneration merit further investigation.

neuroscience↗

bioRxiv2021

Integrative molecular and clinical profiling of acral melanoma identifies LZTR1 as a key tumor promoter and therapeutic target

Acral melanoma, the most common melanoma subtype among non-Caucasian individuals, is associated with poor prognosis. However, its key molecular drivers remain obscure. Here, we performed integrative genomic and clinical profiling of acral melanomas from a cohort of 104 patients treated in North America or China. We found that recurrent, late-arising amplifications of cytoband chr22q11.21 are a leading determinant of inferior survival, strongly associated with metastasis, and linked to downregulation of immunomodulatory genes associated with response to immune checkpoint blockade. Unexpectedly, LZTR1 - a known tumor suppressor in other cancers - is a key candidate oncogene in this cytoband. Silencing of LZTR1 in melanoma cell lines caused apoptotic cell death independent of major hotspot mutations or melanoma subtypes. Conversely, overexpression of LZTR1 in normal human melanocytes initiated processes associated with metastasis, including anchorage-independent growth, formation of spheroids, and increased levels of MAPK and SRC activities. Our results provide new insights into the etiology of acral melanoma and implicate LZTR1 as a key tumor promoter and therapeutic target.

cancer biology↗

bioRxiv2021

Ex vivo mass cytometry analysis reveals a profound myeloid proinflammatory signature in psoriatic arthritis synovial fluid

ObjectivesA number of immune populations have been implicated in psoriatic arthritis (PsA) pathogenesis. This study used mass cytometry (CyTOF) combined with transcriptomic analysis to generate a high-dimensional dataset of matched PsA synovial fluid (SF) and blood leucocytes, with the aim of identifying cytokine production ex vivo in unstimulated lymphoid and myeloid cells. MethodsFresh SF and paired blood were either fixed or incubated with protein transport inhibitors for 6 h. Samples were stained with two CyTOF panels: a phenotyping panel and an intracellular panel, including antibodies to both T cell and myeloid cell secreted proteins. Transcriptomic analysis by gene array of key expanded cell populations and single-cell RNA-seq, and ELISA and LEGENDplex analysis of PsA SF were also performed. ResultsWe observed marked changes in the myeloid compartment of PsA SF relative to blood, with expansion of intermediate monocytes, macrophages and dendritic cell populations. Classical monocytes, intermediate monocytes and macrophages spontaneously produced significant levels of the proinflammatory mediators osteopontin and CCL2 in the absence of any in vitro stimulation. By contrast minimal spontaneous cytokine production by T cells was detected. Gene expression analysis showed the genes for osteopontin and CCL2 to be amongst those most highly upregulated by PsA monocytes/macrophages; and both proteins were elevated in PsA SF. ConclusionsUsing multiomic analyses we have generated a comprehensive cellular map of PsA SF and blood to reveal key expanded myeloid proinflammatory modules in PsA of potential pathogenic and therapeutic importance.

immunology↗

bioRxiv2019

Elevated risk of invasive group A streptococcal disease and host genetic variation in the human leukocyte antigen locus

Invasive group A streptococcal (GAS) disease is uncommon but carries a high case-fatality rate relative to other infectious diseases. Given the ubiquity of mild GAS infections, it remains unclear why healthy individuals will occasionally develop life-threatening infections, raising the possibility of host genetic predisposition. Here, we present the results of a case-control study including 43 invasive GAS cases and 1,540 controls. Using HLA imputation and linear mixed-models, we find each copy of the HLA-DQA1*01:03 allele associates with a two-fold increased risk of disease (odds ratio 2.3, 95% confidence interval 1.3-4.4, P=0.009), an association which persists with classical HLA typing of a subset of cases and analysis with an alternative large control dataset with validated HLA data. Moreover, we propose the association is driven by the allele itself rather than the background haplotype. Overall this finding provides impetus for further investigation of the immunogenetic basis of this devastating bacterial disease.

genetics↗

bioRxiv2018

Objective Evaluation of Multiple Sclerosis Lesion Segmentation using a Data Management and Processing Infrastructure

We present a study of multiple sclerosis segmentation algorithms conducted at the international MICCAI 2016 challenge. This challenge was operated using a new open-science computing infrastructure. This allowed for the automatic and independent evaluation of a large range of algorithms in a fair and completely automatic manner. This computing infrastructure was used to evaluate thirteen methods of MS lesions segmentation, exploring a broad range of state-of-the-art algorithms, against a high-quality database of 53 MS cases coming from four centers following a common definition of the acquisition protocol. Each case was annotated manually by an unprecedented number of seven different experts. Results of the challenge highlighted that automatic algorithms, including the recent machine learning methods (random forests, deep learning, ...), are still trailing human expertise on both detection and delineation criteria. In addition, we demonstrate that computing a statistically robust consensus of the algorithms performs closer to human expertise on one score (segmentation) although still trailing on detection scores.

neuroscience↗

bioRxiv2018

Genetic determinants of risk and survival in pulmonary arterial hypertension

BackgroundPulmonary arterial hypertension (PAH) is a rare disorder leading to premature death. Rare genetic variants contribute to disease etiology but the contribution of common genetic variation to disease risk and outcome remains poorly characterized.\n\nMethodsWe performed two separate genome-wide association studies of PAH using data across 11,744 European-ancestry individuals (including 2,085 patients), one with genotypes from 5,895 whole genome sequences and another with genotyping array data from 5,849 further samples. Cross-validation of loci reaching genome-wide significance was sought by meta-analysis. We functionally annotated associated variants and tested associations with duration of survival.\n\nFindingsA locus at HLA-DPA1/DPB1 within the class II major histocompatibility (MHC) region and a second near SOX17 were significantly associated with PAH. The SOX17 locus contained two independent signals associated with PAH. Functional and epigenomic data indicate that the risk variants near SOX17 alter gene regulation via an enhancer active in endothelial cells. PAH risk variants determined haplotype-specific enhancer activity and CRISPR-inhibition of the enhancer reduced SOX17 expression. Analysis of median survival showed that PAH patients with two copies of the HLA-DPA1/DPB1 risk variant had a two-fold difference (>16 years versus 8 years), compared to patients homozygous for the alternative allele.\n\nInterpretationWe have found that common genetic variation at loci in HLA-DPA1/DPB1 and an enhancer near SOX17 are associated with PAH. Impairment of Sox17 function may be more common in PAH than suggested by rare mutations in SOX17. Allelic variation at HLA-DPB1 stratifies PAH patients for survival following diagnosis, with implications for future therapeutic trial design.\n\nFundingUK NIHR, BHF, UK MRC, Dinosaur Trust, NIH/NHLBI, ERS, EMBO, Wellcome Trust, EU, AHA, ACClinPharm, Netherlands CVRI, Dutch Heart Foundation, Dutch Federation of UMC, Netherlands OHRD and RNAS, German DFG, German BMBF, APH Paris, Inserm, Universite Paris-Sud, and French ANR.

genetics↗

bioRxiv2018

Stratification of amyotrophic lateral sclerosis patients: a crowdsourcing approach

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease with substantial heterogeneity in clinical presentation with an urgent need for better stratification tools for clinical development and care. In this study we used a crowdsourcing approach to address the problem of ALS patient stratification. The DREAM Prize4Life ALS Stratification Challenge was a crowdsourcing initiative using data from >10,000 patients from completed ALS clinical trials and 1479 patients from community-based patient registers. Challenge participants used machine learning and clustering techniques to predict ALS progression and survival. By developing new approaches, the best performing teams were able to predict disease outcomes better than currently available methods. At the same time, the integration of clustering components across methods led to the emergence of distinct consensus clusters, separating patients into four consistent groups, each with its unique predictors for classification. This analysis reveals for the first time the potential of a crowdsourcing approach to uncover covert patient sub-populations, and to accelerate disease understanding and therapeutic development.

bioinformatics↗